Energy metabolism modelling with sensitivity to activity thermogenesis tracking data

这项研究旨在开发更准确的人体能量代谢数学模型。这些模型对于行业合作伙伴Fitmylife Health Analytics非常有价值。该公司的目标是利用这些模式帮助客户建立行为模式，以实现健康目标。这些模型确保客户的决策植根于对自己身体新陈代谢的独特方面的准确理解。人体能量代谢最准确的模型是基于控制人体处理消耗的大量营养素：蛋白质、脂肪和碳水化合物的过程的数学模型来预测身体成分的变化。然而，这些模型存在着严重的缺陷，限制了它们预测任何给定个体的身体成分变化的能力。这些缺陷的例子包括：(A)归因于体力活动水平的单一参数，(B)对建模对象进行的体力活动的类型不敏感，(C)没有关于大量营养素的卡路里摄入量的信息，(D)假定稳态初始条件，以及(E)对一天中的时间卡路里摄入量不敏感。事实上，有可能避免所有这些缺陷，并为模型提供准确的数据。常用的健康跟踪设备提供了由于活动以及活动类型而导致的每分钟心率的准确信息。饮食跟踪应用程序提供按常量营养素和一天中的时间分解卡路里摄入量。有了关于建模对象的历史数据，就有可能避免对稳态初始条件的假设。最后，现在可以用六种成分准确地模拟身体成分，而不是只有两种。通过这个项目，研究小组的目标是开发新陈代谢模型，这些模型受益于关于卡路里摄入量和每日总能量消耗的更准确的信息。最终，我们计划评估这些新模型相对于国家卫生研究所建立的当前领先模型所建立的基线的预测能力。

This research initiative aims to develop more accurate mathematical models of human energy metabolism. These models are of great value to the industrial partner, Fitmylife Health Analytics. The company aims to use the models to help its customers establish behaviour patterns to achieve health goals. The models ensure that customers' decisions are rooted in an accurate understanding of the unique aspects of their own body's metabolism. The most accurate models of human energy metabolism predict body composition change based on mathematical models of processes that govern the body's processing of consumed macronutrients: protein, fat, and carbohydrates. However, these models suffer important deficiencies that limit their power to predict body composition change for any given individual. Examples of these deficiencies include: (a) single parameter ascribed to physical activity level, (b) no sensitivity to the type of physical activity performed by subjects of the modelling, (c) no information on calorie intake by macronutrient, (d) assumption of steady state initial conditions, and (e) no sensitivity to time of day for calorie intake. In fact, it is possible to avoid all of these deficiencies and inform models with precise data. Commonly available fitness tracking devices provide accurate information of minute-by-minute heart rate due to activity along with activity type. Diet tracking applications provide decomposition of calorie intake by macronutrient and time of day. With historical data on modelling subjects it is possible to avoid assumptions of steady state initial conditions. Finally, it is now possible to accurately model body composition with six components rather than just two.Through this project the research team aims to develop models of metabolism that benefit from much more precise information regarding caloric intake and total daily energy expenditure. Ultimately, we plan to evaluate the predictive power of these new models relative to baselines established by the current leading model established by the National Institute of Health.